Google DeepMind has introduced three new additions to its Gemini lineup: Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber. The release focuses on speed, efficiency, and practical use across modern AI workflows.
Gemini 3.6 Flash is positioned as the company's main everyday model, designed to improve coding, knowledge tasks, and multimodal performance while using up to 17% fewer tokens than 3.5 Flash. That efficiency could make it a more economical option for teams building AI products at scale.
Google also highlighted 3.5 Flash-Lite as the most budget-friendly model in the group. Meanwhile, 3.5 Flash Cyber was trained specifically to identify and repair cybersecurity weaknesses, and it will be offered only to governments and selected trusted partners through a limited-access program.
The company says the new models are aimed at delivering stronger reliability, lower latency, and better cost control for developers creating AI agents and production systems. In that sense, the launch reflects a broader shift toward models that are not only powerful, but also practical and scalable.
Notably, the update does not include the long-awaited Gemini Pro refresh. Google had previously indicated that the Pro version was in internal use and expected to arrive soon, while DeepMind product lead Logan Kilpatrick said the team is testing Gemini 3.5 Pro with partners and hopes to release it shortly. He also noted that work has already begun on Gemini 4 pre-training, signaling the next phase of Google's AI roadmap.
As the model race accelerates, Google's latest move suggests a future where AI systems become faster, more specialized, and more efficient across everyday digital work.
